Introduction

Resolving geographic units that do not neatly coincide is a common problem in spatial data analysis. The method outline here attempts to conflate King County Health Reporting Areas (HRAs) to US Census tracts. In the cases where a given tract is entirely within an HRA, that tract receives the HRA’s unique identifier (HRA_ID). On the other hand, in cases where a given tract overlaps multiple HRAs, block-level census data is used to determine which HRA ID to assign to the tract.

Census Block Counts

This method provides three alternatives of block-level counts that can be used:

Count Type	Variable ID	Source
Population	POP	Table P1, U.S. Census Bureau, 2010 Census
Housing Units	HU	Table H1, U.S. Census Bureau, 2010 Census
Population in Housing Units	HUPOP	Table H10, U.S. Census Bureau, 2010 Census

The Algorithm

The following actions are performed in this method:

1. Centroids of the census block polygons are calculated (class = SpatialPointsDataFrame)
2. HRA IDs are passed to the block centroid using a spatial overlay method (sp::over())
3. Blocks are aggregated into tracts and the count variables (POP,HU,POPHU) are summed
4. For each count variable, the HRA ID with the highest sum is assigned to each tract

After running the assignment algorithm, it is clear that the POP and POPHU variables result in the same HRA assignments. HU differs from the other two variables in only 3 of the 398 tracts:

GEOID_TR	HRA_POP	HRA_POPHU	HRA_HU
53033022202	Kirkland North	Kirkland North	Kirkland
53033025001	Bellevue-South	Bellevue-South	Newcastle/Four Creeks
53033028801	SeaTac/Tukwila	SeaTac/Tukwila	Des Moines/Normandy Park

Maps